Controls for Production (#9)
/In this episode, we discuss the controls associated with production. A major concern in this area is that the production staff is in the business of manufacturing goods, so any controls imposed on them will reduce their production efficiency. Consequently, controls generally need to be built into the production system, with little user interaction, or require the services of a separate group of people from those engaged in production. Key points made in the podcast are noted below.
The Need to Minimize Controls
Today’s topic is controls for production. Now, production employees are in the business of building products, so if you ask them to do anything else, like fill out forms or count things, they’ll be distracted. The result is that the controls are performed poorly, and because the employees are distracted, this may also have a negative impact on their regular production work. Therefore, it’s best to install production controls that don’t bother the production staff. This can mean that control work is only done by non-production employees, or that controls are installed in processes around the production area, but not within it.
What to Control in Production
What are we controlling in production? We want to make sure that the right orders are being produced, and that the materials and staff are available to manufacture those orders. To ensure that the right order is produced, the standard control is to issue work orders from a master production schedule, which is maintained by the production planning staff. A production routing instruction sheet will be attached to the initial parts kit when it departs the warehouse, and the materials handler will shift it from work station to work station in accordance with the work sequence listed on the routing instruction.
The production planning staff monitors the flow of orders through the production process. Though there are some automated methods available to conduct the monitoring, the most common approach is to have the production planners tour the plant and manually verify the status of each order. Conversely, the automated approaches are to either have each workstation operator log in the arrival and departure of each production job on a computer monitor, or RFID tags can be attached to each order, and monitored based on the signals the tags emit. The use of RFID tags is still exploratory, so I don’t recommend that approach until the technology has settled down a bit more.
Ideally, all the parts are used, production is completed, and the resulting finished goods are returned to the warehouse.
Unfortunately, it’s not quite that simple. We’re still going to have problems ensuring that the correct materials and staff are available to produce those goods.
One problem at the beginning of the production process is that the picking staff in the warehouse may not have properly picked all of the parts. To detect this, a logical control is to have the materials handler compare the picked parts to the parts listed on the pick list, to make sure that they match.
Next, what if the warehouse staff picked exactly what was listed on the pick list, but it was not what the production staff really needed? This problem is caused by an incorrect bill of materials. There are several ways to control this issue. First, lock down access to the bill of material file in the computer system, so that only authorized people can access it. Second, audit the bills of material, and have all problems corrected. The target bill of material accuracy level, by the way, should be at least 98%. And third, investigate whenever the production staff has to request extra parts from the warehouse, or if they return unused parts to it – both are indicators of an incorrect bill of materials.
A different problem that can occur anywhere in the production process is that the routing document is incorrect, and doesn’t properly include all production work steps. This is important, since the problem jobs must be jammed into the backlog of work stations that are not even expecting the work, which can cause some major backlogs. To fix this problem, audit the labor routing records, and shoot for a target accuracy level of at least 95%.
Also, as was the case for bills of material, always lock down access to the labor routing records in the computer, so that only an authorized person can access them.
Yet another problem is the occasional production of scrap. An allowance for scrap is sometimes included in the bill of materials, but if the allowance is too small, then the job will run out of materials before the required number of units have been completed. There are several ways to ensure that scrap levels are correctly recorded.
One is to have the production staff fill out a form that itemizes how much of which item was scrapped. Then a data entry person collects the forms and enters the scrap information into the computer.
It also helps to prenumber the scrap forms and track down any missing forms. However, as I mentioned earlier, you generally want to keep paperwork away from the production staff. Also, they have a tendency not to report scrap, because they don’t want to be held responsible for making the scrap. So, we need some alternative ways to collect scrap information. One is to have the production staff dump all scrap into collection bins, and then a data entry person adds it all up at the end of the shift and records the amount of scrap. A less organized approach is to periodically sweep through the production area and collect any scrap or finished goods that are left lying around, and then record this information.
There are other issues that need controls, such as labeling goods for rework, or pulling inventory out of production for a quality review, but you get the general idea that unusual transactions that remove materials from the production flow must be reported as fast as possible.
Material Requirements Planning as a Control
For those of you with some experience in manufacturing, the production flow that I’ve been talking about is based on a material requirements planning system, or MRP for short. An MRP system is a push system, whereby a production schedule plans out how many units will be produced, and what materials will be required to produce it, and then controls the entire process of ordering materials and shepherding them through the manufacturing process.
Now, the problem with an MRP system is that it requires massive amounts of accurate information in order to correctly plan the production process. To ensure that the MRP system works, there need to be lots of controls to verify that the data entering the system does not screw up the works.
Just-in-Time Systems and Production Controls
A way to reduce the number of controls is to use a just-in-time system. This approach is based on the pull concept, which is that finished goods are only manufactured when there is actual demand for it. A signaling device called a kanban is used to notify upstream workstations of the quantity needed, and those workstations produce until the order represented by the kanban has been filled. Then they stop working. Some other features of a just-in-time system include very small lot sizes, short lead times, frequent supplier deliveries, and a strong focus on reducing equipment setup times.
The just-in-time system is very attractive from a controls perspective.
First of all, lot sizes tend to be quite small, so scrap is discovered by the downstream workstation almost at once, before the upstream workstation has had much time to create more of it. This means that scrap levels will be minimal, and so there’s not much need for a control to count scrap.
Second, there’s little need to count incoming inventory, either from the warehouse or from an upstream workstation, because it’s usually slotted into bays in a standard-sized container that’s easy to count with just a glance. So, this eliminates the need for a quantity verification control.
Third, work-in-process inventory doesn’t stay in the manufacturing area for very long, so there’s no need to track inventory as it moves from workstation to workstation within the production area. Instead, just count the amount of finished goods as they come out of the production area.
The main area in which controls are needed for a just-in-time system involves suppliers. They’ll be making lots of small-quantity deliveries, possibly several per day, and it’s not efficient for the receiving staff to inspect and log in each receipt. Instead, the location of controls need to shift upstream to the supplier itself, where the company’s industrial engineering staff pre-certifies the ability of each supplier to deliver the right quantities, on time, with good quality, and directly to the production area. This may also call for the use of a supplier performance scoring system.
In short, a just-in-time manufacturing system requires far less control than a traditional MRP system.
Parting Thoughts
Here’s my view of production controls: The overriding concern in production is to stay out of the way of the production staff. Their job is to manufacture goods. This means that any attempt to have them fill out forms, conduct counts, or do anything even remotely associated with a control will almost certainly reduce their level of efficiency. Unfortunately, an MRP system requires a lot of paperwork – it needs to be notified of any transactions that are even remotely out of the ordinary. On the other hand, a just-in-time system is designed to be essentially self-controlling. And that means that discrete control functions are minimized, and the production staff can get on with the business of production. Obviously, I like the control environment for just-in-time manufacturing.